One of the most common Android malware detection methods is the use of the static analysis method since it analyzes the internal architecture of an APK file without running it. This is done by scanning permissions, manifest attributes, code instruction, and API invocation to determine that there are evil patterns within the application. Machine learning models can effectively identify benign and malicious apps by transforming extracted features into numbers. Compared to dynamic execution, static analysis is safer and faster hence can be used in large scales to screen malware. Its accuracy however decreases in the presence of highly obfuscated code, encrypted payloads or malware that only triggers malicious behaviour once it is in execution.
Keywords
^KEYWORDS^
Conclusion
^CONCLUSION_TEXT^
References
Android Malware
Detection using Static Analysis (2021) Li Wang, M. Kumar, Banerjee Fast analysis; no execution required; low resource usage; suitable for large-scale screening. Ineffective against obfuscated or packed malware; cannot
detect runtime
behaviour; limited zero-day detection.
Dynamic Behaviour- Based Android Malware Detection (2022) R. Sharma, D. Patel, Thomas Captures real execution traces; resistant to obfuscation; reveals hidden payloads; strong for spyware and
ransomware. High computational overhead; slow execution; vulnerable to sandbox-evasion; requires realistic input
simulation.
Hybrid Static–Dynamic Android Malware Detection (2023) H. Zhao, S. Mehta, Vivek Rao High detection accuracy; reduced false positives; strong zero-day detection; resilience against code hiding. Complex architecture; high hardware requirement; difficult on-device
deployment; requires
large datasets. Deep Learning–Driven Android Malware Classification (2024) A.Gupta, F. Hernandez,
Y. Cho Learns deep patterns automatically; very high accuracy; effective for malware families and
zero-day attacks. Requires GPUs/TPUs; large dataset dependency; limited model interpretability;
prone to overfitting.
Network Traffic Analysis for Android Malware (2021) J. Singh, R. Alami, O. Musa Effective for botnets and C2 detection; low device resource usage; independent of code structure. Encrypted traffic evades detection; high false positives; requires continuous monitoring; cannot
detect offline threats.
Machine Learning- Based APK
Classification (2020) Priya Nair, Sohail Khan, Ankit Verma Automated classification; scalable; integrates static and dynamic features; supports family-level grouping. Feature dependency; frequent retraining required; vulnerable to adversarial
manipulation; dataset imbalance issues.
How to Cite This Paper
Sandhya S, Preethi CS, Preethi JL,
Dhanushree HT (2025). Android Malware Detection From APK File. International Journal of Computer Techniques, 12(6). ISSN: 2394-2231.
